“Languages are deeply linked to lives, to cultures, to ethnic groups and political organizations.”
Roberto Zariquiey, Linguist
Pontifical Catholic University, Peru
Spoken by more than 300 million people in dozens of countries all over the world, these 24 newcomers bring the total Google Translate portfolio to 133 languages. Used mainly in Africa, South Asia, and South America, this is the largest batch since sixteen new languages were added in 2010. Five more languages were added to the platform in 2020.
Underpinned by what Google calls a zero-resource or zero-shot machine translation model, this innovative platform can learn to translate into other languages without ever seeing written examples.
This is a major step forward for languages with oral traditions, particularly in the fields of healthcare and education. In rural areas, practitioners and educators (qualified in mainstream institutions) often struggle to communicate with locals speaking only indigenous languages.
How Does Google Translate Work?
In simple terms, GT works through training. Around a hundred data-rich languages are fed into a single enormous artificial intelligence engine (the platform’s ‘brain’), which applies all this knowledge to dozens of other languages. It then starts to work out what words and phrases might mean, based on its knowledge of multiple other languages.
However, Google itself agrees that this output is still far from perfect―especially when it comes to minority languages. For GT research scientist Isaac Caswell, machine translation is great for conveying ideas quickly and cheaply, although losing much of the subtlety of the language in the process―and sometimes much of its meaning as well!
Why Are These 24 Languages Important?
From missionaries to medics and marketers, reaching out to audiences in their own languages has long been a no-brainer for communicators. However, cultural factors have raised hurdles. Although Google Translate has long been available in minority European languages (Icelandic, Frisian, Corsican, and Maltese) with fewer than a million speakers each, East African languages with millions of speakers (Oromo and Tigrinya) have been neglected.
This latest batch of languages makes a good start on reversing this situation, helping millions of people whose mother tongues are poorly represented in today’s technology. Intended as stepping-stones for communities struggling to keep pace with the modern world, these latest additions are clustered on three continents:
- Africa: Bambara. Ewe, Krio, Lingala, Luganda, Manipuri Meiteilon, Oromo, Sorani Kurdish, Tigrinya, Sepedi, Tsonga, and Twi
- South Asia: Assamese, Sanskrit, Bhojpuri, Dhivehi, Dogri, Ilocano, Konkani, Maithili, and Mizo
- South America: Aymara, Guarani, and Quechua
The sizes of these language communities range from the compact Dhivehi community of around 300,000 in the Maldives to fifty million Bhojpuri speakers in Nepal, Fiji and Northern India, with almost as many (45 million) speaking Lingala across Central Africa.
Strengthening Communities by Streamlining Communications
Determined to connect communities by breaking down the language barriers that keep people apart, this U.S.-based tech giant views these new additions as far more than just a technological milestone. Google’s latest I/O translation platform is breaking fresh ground in building links with huge but underserved language communities―like ten million Quechua speakers in South America.
Once the lingua franca of the thriving Inca Empire―whose territory stretched from what is now central Chile to southern Colombia―Quechua declined after Spanish conquistadores swept through what Europe called the New World. Nevertheless, it’s still an official language (alongside Aymara) in Peru and Bolivia.
Machine Translation: The First Step in Your Communications Journey
For its many users, machine translation offers two unbeatable advantages over its human counterparts: it’s instant, and it’s free!
However, the quality of raw MT text is still far from acceptable. This is why many clients are meeting their translation needs through a time-saving and cost-efficient hybrid solution: Machine Translation Post-Editing (MTPE).
Offering the best of both worlds, skilled MTPE editors can halve the time needed by human translators, while upgrading MT text quality and accuracy to the human level.
How Do I Decide Between Human Translation and MTPE?
When it comes to translations, there’s no one-size-fits-all solution. Every document is unique, with its own text and timeframe, and must be assessed individually. The most efficient solution is a customized quote from the Trusted Translations experts, tailored to your budgets and deadlines.
Photo by Jess Bailey on Unsplash